1015367.pdf

Research Article
Optimal Placement of Wind Power Plants in Transmission Power
Networks by Applying an Effectively Proposed
Metaheuristic Algorithm
Minh Quan Duong ,1
Thang Trung Nguyen ,2
and Thuan Thanh Nguyen 3
1
Department of Electrical Engineering, University of Science and Technology-The University of DaNang,
Da Nang 55000, Vietnam
2
Power System Optimization Research Group, Faculty of Electrical and Electronics Engineering, Ton Duc Thang University,
Ho Chi Minh City 700000, Vietnam
3
Faculty of Electrical Engineering Technology, Industrial University of Ho Chi Minh City, Ho Chi Minh City 700000, Vietnam
Correspondence should be addressed to Thang Trung Nguyen; nguyentrungthang@tdtu.edu.vn
Received 7 September 2021; Revised 28 September 2021; Accepted 29 September 2021; Published 12 October 2021
Academic Editor: Yang Li
Copyright © 2021 Minh Quan Duong et al. This is an open access article distributed under the Creative Commons Attribution
License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is
properly cited.
In this paper, a modified equilibrium algorithm (MEA) is proposed for optimally determining the position and capacity of wind
power plants added in a transmission power network with 30 nodes and effectively selecting operation parameters for other
electric components of the network. Two single objectives are separately optimized, including generation cost and active power
loss for the case of placing one wind power plant (WPP) and two wind power plants (WPPs) at predetermined nodes and
unknown nodes. In addition to the proposed MEA, the conventional equilibrium algorithm (CEA), heap-based optimizer (HBO),
forensic-based investigation (FBI), and modified social group optimization (MSGO) are also implemented for the cases. Result
comparisons indicate that the generation cost and power loss can be reduced effectively, thanks to the suitable location selection
and appropriate power determination for WPPs. In addition, the generation cost and loss of the proposed MEA are also less than
those from other compared methods. Thus, it is recommended that WPPs should be placed in power systems to reduce cost and
loss, and MEA is a powerful method for the placement of wind power plants in power systems.
1. Introduction
Solving optimal power flow problem (OPF) to have the
steady and effective states of power systems is considered as
the leading priority in operation of power systems. Specif-
ically, the steady state is represented as a state vector and
regarded as a set of variables, such as output of active and
reactive power from power plants, voltage of generators in
power plants, output of reactive power from shunt capac-
itors, transformers’ tap, voltage of loads, and operating
current of transmission lines [1–3]. Generally, during the
whole process of solving the OPF problem to determine the
steady state in power system operation, the mentioned
variables are separated into control variables and dependent
variables [4, 5]. The output of the reactive power from power
plants (QG), output of the active power of power plants at
slack node (PGs), voltage of loads (VL), and current of lines
(Il) are grouped in a dependent variable set [6–10], whereas
other remaining variables including tap changer of trans-
formers (TapT), output of the active power from the gen-
erators excluding that at slack node (PGs), and output of the
reactive power supplied by capacitor banks (QCap) are put in
a control variable set [11–15]. These control variables are
utilized as the input of the Mathpower programme to find
the dependent variables. The Mathpower programme is a
calculating tool developed based on the Newton–Raphson
method to deal with power flow. After having the dependent
variable set, it is checked and penalized based on previously
known upper bound and lower bound. The violation of the
bounds will be considered for the quality of both control and
Hindawi
Mathematical Problems in Engineering
Volume 2021,Article ID 1015367, 20 pages
https://doi.org/10.1155/2021/1015367

dependent variable sets [16–20]. These violations are con-
verted into penalty terms and added to objective functions,
such as electrical power generation cost (Coste), active power
loss (Ploss), polluted emission (Em), and load voltage stability
index (IDsl).
Recently, the presence of renewable energies has been
considered in power systems when the percentages of wind
power and solar energy joining into the process of gener-
ating electricity become more and more. In that situation,
the OPF problem was modified and became more complex
than ever. The conventional version of the OPF problem
only considers thermal power plants (THPs) as the main
source [21–24]. Other modified versions of the OPF prob-
lem, both THPs and renewable energies, are power sources.
The modified OPF problem is outlined in Figure 1 in which
the conventional OPF problem is a part of the figure without
variables regarding renewable energies, such as output of
active and reactive power of wind power plant (Pw, Qw),
output of active and reactive power of photovoltaic power
plants (PVPs) (Ppv, Qpv), and location of WPPs and PVPs
(Lw, Lpv). There are large number of studies proposed to
handle the modified OPF problems. These studies can be
classified into three main groups. Specifically, the first group
solves the OPF problem considering wind power source
injecting both active and reactive power into grid. The
second group considers the assumption that wind energy
sources just generate active power only. The third group
considers both wind and solar energies in the process of
solving the OPF problem. The applied methods, test systems,
objective functions, placed renewable power plants, and
compared methods regarding modified OPF problems are
summarized in Table 1. All the studies in the table have
focused on the placement of wind and photovoltaic power
plants to cut electricity generation fuel cost for THPs, and
the results were mainly compared to base systems without
the contribution of the renewable plants. In addition, other
research directions of optimal power flow are without re-
newable power plants but using reactive power dispatch
[50, 51] and using VSC (voltage source converter) based on
HVDC (high-voltage direct current) [52, 53]. These studies
also achieved the reduction of cost and improved the quality
of voltage as expected. If the combination of both using
renewable energies and optimal dispatch of reactive power
or the combination of using both renewable energies and
these converters can be implemented, expected results such
as the reduction of cost and power loss and the voltage
enhancement can be significantly better.
In recent years, metaheuristic algorithms have been
developed widely and applied successfully for optimization
problems in engineering. One of the most well-known al-
gorithms is the conventional equilibrium algorithm (CEA)
[54], which was introduced in the early 2020. The con-
ventional version was demonstrated more effective than
PSO, GWA, GA, GSA, and SSA for a set of fifty-eight
mathematical functions with a different number of variables
and types. Over the past year and this year, CEA was widely
replicated for different optimization problems such as AC/
DC power grids [55], loss reduction of distribution networks
[56], component design for vehicles [57], and multidisci-
plinary problem design [58]. However, the performance of
CEA is not the most effective among utilized methods for the
same problems. Consequently, CEA had been indicated to
be effective for large-scale problems, and it needs more
improvements [59–62]. Thus, we proposed another version
of CEA, called the modified equilibrium algorithm (MEA),
and also applied four other metaheuristic algorithms for
checking the performance of MEA.
In this paper, the authors solve a modified OPF
(MOPF) problem with the placement of wind power
plants in an IEEE 30-bus transmission power network.
About the number of wind power plants located in the
system, two cases are, respectively, one wind power plant
(WPP) and two WPPs. About the locations of the WPPs,
simple cases are referred to the previous study [47] and
other more complicated cases are to determine suitable
buses in the system by applying metaheuristic algorithms.
It is noted that the study in [47] has only studied the
placement of one WPP, and it has indicated the most
suitable bus as bus 30 and the most ineffective bus as bus 3.
In this paper, we have employed buses 3 and 30 for two
separated cases to check the indication of the study [47].
The results indicated that the placement of one WPP at
bus 30 can reach smaller power loss and smaller fuel cost
than at bus 3. In addition, the paper also investigated the
effectiveness of locations by applying MEA and four other
metaheuristic algorithms to determine the location. As a
result, placing one WPP at bus 30 has reached the smallest
power loss and the smallest total fuel cost. For the case of
placing two WPPs, buses 30 and 3 could not result in the
smallest fuel cost and the smallest power loss. Buses 30
and 5 are the best locations for the minimization of fuel
cost, while buses 30 and 24 are the best locations for the
minimization of power loss. Therefore, the main contri-
bution of the study regarding the electrical field is de-
termining the best locations for the best power loss and
the best total cost.
All study cases explained above are implemented by the
proposed MEA and four existing algorithms published in
2020, including conventional equilibrium algorithm (CEA)
[54], heap-based optimizer (HBO) [63], forensic-based in-
vestigation (FBI) [64], and modified social group optimi-
zation (MSGO) [65]. As a result, the best locations leading to
the smallest cost and smallest loss are obtained by MEA.
Thus, the applications of the four recent algorithms and the
proposed MEA aim to introduce a new algorithm and show
their effectiveness to readers in solving the MOPF problem.
Readers can give evaluations and decide if the algorithms are
used for their own optimization problems, which maybe in
electrical engineering or other fields. The major contribu-
tions of the paper are summarized again as follows:
(1) Find the best locations for placing wind power plants
in the IEEE 30-bus transmission power gird.
(2) The added wind power plants and other found pa-
rameters of the system found by MEA can reach the
smallest power loss and smallest total cost.
2 Mathematical Problems in Engineering

(3) Introduce four existing algorithms developed in 2020
and a proposed MEA. In addition, the performance
of these optimization tools is shown to readers for
deciding if these tools are used for their applications.
(4) Provide MEA, the most effective algorithm among
five applied optimization tools for the MOPF
problem.
The organization of the paper is as follows. Two single
objectives and a considered constraint set are presented in
Section 2. The configuration of CEA for solving a sample
optimization problem and then modified points of MEA are
clarified in detail in Section 3. Section 4 summarizes the
computation steps for solving the modified OPF problem by
using MEA. Section 5 presents results obtained by the
proposed MEA and other methods such as JYA, FBI, HBO,
and MSGO. Finally, conclusions are given in Section 6 for
stating achievements in the paper.
2. Objective Functions and Constraints of the
Modified OPF Problem
2.1. Objective Functions
2.1.1. Minimization of Electricity Generation Cost. In this
research, the first single objective is considered to be elec-
tricity generation cost of all thermal generators. At generator
nodes, where thermal units are working, the cost is the most
Run
power
flow
Ppv Qpv
Lpv
PG
VG
Tapt
Qcap
Pw Qw
Lw
PGs
QG
Il
Vl
Coste
Ploss
IDsl
Em
Figure 1: Configuration of the modified OPF problem in the presence of renewable energies.
Table 1: The summary of studies proposed to solve the modified OPF problem considering renewable energies.
Reference Method Applied system Renewable energy Compared methods
[25] BFA 30 nodes Wind (P, Q) GA
[26] MBFA 30 nodes Wind (P, Q) ACA
[27] HABC 30 nodes Wind (P, Q) ABA
[28] MCS 30, 57 nodes Wind (P, Q) MPSO
[29] HA 30 nodes Wind (P, Q) PSO
[30] MFO 30 nodes Wind, solar (P, Q) GWA, MVA, IMA
[31] AFAPA 30, 75 nodes Wind (P) APO, BA
[32] KHA 30, 57 nodes Wind (P) ALPSO, DE, RCGA
[33] GSO 300 nodes Wind (P) NSGA-II
[34] MHGSPSO 30, 57 nodes Wind (P, Q) MSA, GWA, WA
[35] BSA 30 nodes Wind, solar (P) —
[36] IMVA 30 nodes Wind, solar (P, Q) PSO, MVA, NSGA-II
[37] PSO 39 nodes Wind, solar (P) PSO variants
[38] NSGA-II 30, 118 nodes Wind, solar (P) —
[39] MFA 30 nodes Wind, solar (P, Q) MDE
[40] GWO 30, 57 nodes Wind, solar (P) GA, PSO, CSA1, MDE, ABA
[41]
BWOA, ALO, PSO
30 nodes Wind, solar (P, Q) —
GSA, MFA, BMA
[42] FPA IEEE 30-bus Wind, solar (P) —
[43] APDE IEEE 30-bus Wind, solar (P, Q) —
[44] HGTPEA 30 nodes Wind, solar (P) —
[45] HGNIPA 118 nodes Wind, solar (P, Q) —
[46] NDSGWA 30 nodes Wind, solar (P) MDE
[47] JYA 30 nodes Wind, solar (P) —
[48] HSQTIICA 30 nodes DG (P, Q) IICA
[49] MJYA 30, 118 nodes Wind (P) MSA, ABA, CSA, GWA, BSA
Mathematical Problems in Engineering 3

important factor in optimal operation of the distribution
power networks, and it should be low reasonably as the
following model. The total cost is formulated by
EGC � 􏽘
NTG
i�1
FETGi PTGi􏼁, (1)
where FETGi is the fuel cost of the ith thermal unit and
calculated as follows:
FETGi PTGi􏼁 � μ1i + μ2iPTGi + μ3i PTGi􏼁
2
. (2)
2.1.2. Minimization of Active Power Loss. Minimizing active
power loss (APL) is a highly important target in trans-
mission line operation. In general, reactive power loss of
transmission power networks is very significant due to a
high number of transmission lines with high operating
current. If the loss can be minimized, the energy loss and the
energy loss cost are also reduced accordingly. The loss can be
obtained by different ways as follows:
APL � 􏽘
NBr
q�1
3.I2
q.Rq,
APL � 􏽘
NTG
i�1
PTGi − 􏽘
Nn
i�1
Prqi,
APL � 􏽘
Nn
x�1
􏽘
Nn
y�1,
x ≠ y
Yxy U2
x + V2
y − 2UxUy cos φx − φy
􏼐 􏼑
􏽨 􏽩.
(3)
2.2. Constraints
2.2.1. Physical Constraints regarding Thermal Generators.
In the operating process of thermal generators, three main
constraints need to be supervised strictly consisting of the
limitation of real power output, the limitation of reactive
power output, and the limitation of the voltage magnitude.
The violation of any limitations as mentioned will cause
damage and insecure status in whole system substantially.
Thus, the following constraints should be satisfied all the
time:
Pmin
TGi ≤ PTGi ≤ Pmax
TGi , with i � 1, . . . , NTG,
Qmin
TGi ≤ QTGi ≤ Qmax
TGi , with i � 1, . . . , NTG,
Umin
TGi ≤ UTGi ≤ Umax
TGi , with i � 1, . . . , NTG.
(4)
2.2.2. The Power Balance Constraint. The power balance
constraint is the relationship between source side and
consumption side in which sources are TUs and renewable
energies, and consumption side is comprised of loads and
loss on lines. The balance status is established when the
amount of power supplied by thermal generators equals to
the amount of power required by load plus the loss.
Active power equation at each node x is formulated as
follows:
PTGx − Prqx � Ux 􏽘
Nn
y�1
Uy Yxy cos φx − φy
􏼐 􏼑 + Xxy sin φx − φy
􏼐 􏼑
􏽨 􏽩.
(5)
For the case that wind turbines supply electricity at node
x, the balance of the active power is as follows:
PTGx + PWindx − Prqx � Ux 􏽘
Nn
y�1
Uy Yxy cos φx − φy
􏼐 􏼑 + Xxy sin φx − φy
􏼐 􏼑
􏽨 􏽩, (6)
where PWindx is the power generation of wind turbines at
node x and limited by the following constraint:
Pmin
Wind ≤ PWindx ≤ Pmax
Wind. (7)
Similarly, reactive power is also balanced at node x as the
following model:
QTGx + QComx − Qrqx � Ux 􏽘
Nn
y�1
Uy Yxy sin φx − φy
􏼐 􏼑 − Xxy cos φx − φy
􏼐 􏼑
􏽨 􏽩, (8)
where
Qmin
Comx ≤ QComx ≤ Qmax
Comx, with i � 1, . . . , NCom. (9)
For the case that wind turbines are placed at node x, the
reactive power is also supplied by the turbine as the role of
thermal generators. As a result, the reactive power balance is
as follows:

QTGx + QWindx + QComx − Qrqx � Ux 􏽘
Nn
y�1
Uy Yxy sin φx − φy
􏼐 􏼑 − Xxy cos φx − φy
􏼐 􏼑
􏽨 􏽩, ∈ (10)
where QWindx is the reactive power generation of wind
turbines at node x and is subject to the following constraint:
Qmin
Wind ≤ QWindx ≤ Qmax
Wind. (11)
2.2.3. Other Inequality Constraints. These constraints are
related to operating limits of electric components such as
lines, loads, and transformers. Lines and loads are de-
pendent on other operating parameters of other compo-
nents like TUs, wind turbines, shunt capacitors, and
transformers. However, operating values of lines and loads
are very important for a stable operating status of networks.
If the components are working beyond their allowable
range, networks are working unstably, and fault can occur
in the next phenomenon. Thus, the operating parameters of
loads and lines must be satisfied as shown in the following
models:
Umin
LN ≤ ULNt ≤ Umax
LN , with t � 1, . . . , NLN,
SBrq ≤ Smax
Br , with q � 1, . . . , NBr.
(12)
In addition, transformers located at some nodes need to
be tuned for supplying standard voltage within a working
range. The voltage regulation is performed by setting tap of
transformers satisfying the following constraint:
Tapmin
≤ Tapi ≤ Tapmax
, with i � 1, . . . , NT. (13)
3. The Proposed Modified Equilibrium
Algorithm (MEA)
3.1. Conventional Equilibrium Algorithm (CEA). CEA was
first introduced and applied in 2020 for solving a high
number of benchmark functions. The method was superior
to popular and well-known metaheuristic algorithms, but its
feature is simple with one technique of newly updating
solutions and one technique of keeping promising solutions
between new and old solutions.
The implementation of CEA for a general optimization
problem is mathematically presented as follows.
3.1.1. The Generation of Initial Population. CEA has a set of
N1 candidate solutions similar to other metaheuristic al-
gorithms. The solution set needs to define the boundaries in
advance, and then it must be produced in the second stage.
The set of solution is represented by Z � [Zd], where d � 1,
. . ., N1, and the fitness function of the solution set is rep-
resented by Fit � [Fitd], where d � 1, . . ., N1.
To produce an initial solution set, control variables in-
cluded in each solution and their boundaries must be, re-
spectively, predetermined as follows:
Zd � zj d
􏽨 􏽩; j � 1, . . . , N2; d � 1, . . . N1,
Zlow � zjlow
􏽨 􏽩; j � 1, . . . , N2,
Zup � zjup
􏽨 􏽩; j � 1, . . . , N2,
(14)
where N2 is the control variable number, zjd is the jth
variable of the dth solution, Zlow and Zup are lower and upper
bounds of all solutions, respectively, and zjlow and zjup are the
minimum and maximum values of the jth control variable,
respectively.
The initial solutions are produced within their bounds
Zlow and Zup as follows:
Zd � Zlow + r1 · Zup − Zlow
􏼐 􏼑; d � 1, . . . , N1. (15)
3.1.2. New Update Technique for Variables. The matrix fit-
ness fit is sorted to select the four best solutions with the
lowest fitness values among the available set. The solution
with the lowest fitness is set to Zb1, while the second, third,
and fourth best solutions with the second, third, and fourth
lowest fitness functions are assigned to Zb2, Zb3, and Zb4. In
addition, another solution, which is called the middle so-
lution (Zmid) of the four best solutions, is also produced by
Zmid �
Zb1 + Zb2 + Zb3 + Zb4􏼁
4
. (16)
The four best solutions and the middle solution are
grouped into the solution set Z5b as follows:
Z5b � Zb1, Zb2, Zb3, Zb4, Zmid
􏼈 􏼉. (17)
As a result, the new solution Zdnew of the old solution Zd
is determined as follows:
Zdnew � Z5brd + M · Zd − Z5brd􏼁 +
K × M
r2
(1 − M).
(18)
In the above equation, Z5brd is a randomly chosen so-
lution among five solutions of Z5b in equation (17), whereas
M and K are calculated by
M � 2sign r3 − 0.5􏼁
1
eA.Iter
− 1
􏼠 􏼡, (19)
K � K0 · Z5brd − r4 · Zd􏼁, (20)
A � 1 −
Iter
N3
􏼠 􏼡
Iter/N3
( )
, (21)
K0 �
0, if & r5 < 0.5,
r6
2
, & otherwise.
⎧
⎪
⎪
⎨
⎪
⎪
⎩
(22)

3.1.3. New Solution Correction. The new solution Zdnew is a
set of new control variables zjd,,new that can be beyond the
minimum and maximum values of control variables. It
means zjd,,new may be either higher than zjup or smaller than
zjlow. If one out of the two cases happens, each new variable
zjd,,new must be redetermined as follows:
zj d,new �
zjd,new, if zjlow ≤ zjd,new ≤ zjup,
zjlow, if zjlow > zjd,new,
zjup, if zjd,new > zjup.
⎧
⎪
⎪
⎪
⎨
⎪
⎪
⎪
⎩
(23)
After correcting the new solutions, the new fitness
function is calculated and assigned to Fitdnew.
3.1.4. Selection of Good Solutions. Currently, there are two
solution sets, one old set and one new set. Therefore, it is
important to retain higher quality solutions so that the
retained solutions are equal to N1. This task is accomplished
by using the following formula:
Zd �
Zdnew, if Fitd > Fitdnew,
Zdnew, if Fitd � Fitdnew,
Zd, else,
⎧
⎪
⎪
⎨
⎪
⎪
⎩
(24)
Fitd �
Fitdnew, if Fitd > Fitdnew,
Fitdnew, if Fitd � Fitdnew,
Fitd, else.
⎧
⎪
⎪
⎨
⎪
⎪
⎩
(25)
3.1.5. Termination Condition. CEA will stop updating new
control variables when the computation iteration reaches the
maximum value N3. In addition, the best solution and its
fitness are also reported.
3.2. The Proposed MEA. The proposed MEA is a modified
variant of CEA by using a new technique for updating new
control variables. From equation (18), it sees that CEA only
chooses search spaces around the four best solutions and the
middle solution (i.e., Zb1, Zb2, Zb3, Zb4, Zmid) for updating
decision variables whilst from search spaces nearby from the
fifth best solution to the worst solution are skipped inten-
tionally. In addition, the strategy has led to the success of
CEA with better performance than other metaheuristics.
However, CEA cannot reach a higher performance because
it is coping with two shortcomings as follows:
(1) The first shortcoming is to pick up one out of five
solutions in the set Z5b randomly. The search spaces
may be repeated more than once and even too many
times. Therefore, promising search spaces can be
exploited ineffectively or skipped unfortunately.
(2) The second shortcoming is to use two update steps
including M × (Zb1 − Zd) and (K × M/r2)(1 − M),
which are decreased when the computation iteration
is increased. Especially, the steps become zero at final
computation iterations. In fact, parameter A in
equation (21) becomes 0 when the current iteration is
equal to the maximum iteration N3. If we substitute
A � 0 into equation (19), M becomes 0.
Thus, the proposed MEA is reformed to eliminate the
above drawbacks of CEA and reach better results as solving
the OPF problem with the presence of wind energy. The two
proposed formulas for updating new decision variables are
as follows:
Zdnew1 � Zd + M Zb1 − Zd􏼁 + r7 Zr1 − Zr2􏼁, (26)
Zdnew2 � Zb1 + M Zd − Zb1􏼁 + r8 Zr3 − Zr4􏼁. (27)
The two equations above are not applied simultaneously
for the same old solution i. Either equation (26) or equation
(27) is used for the dth new solution. Zdnew1 in equation (26)
is applied to update Zd if Zd has better fitness than the
medium fitness of the population, i.e., Fitd < Fitmean. For the
other case, i.e., Fitd ≥ Fitmean, Zdnew2 in equation (27) is
determined.
4. The Application of the Proposed MEA for
OPF Problem
4.1. Generation of Initial Population. The problem of placing
wind turbines in the transmission power network is suc-
cessfully solved by using decision variables as follows: the
active power generation and voltage of thermal generators
(excluding power generation at slack node), generation of
capacitors, tap of transformer, and position, active and
reactive power of wind turbines. Hence, Zd is comprised of
the following decision variables: PTGi (i � 2, . . ., NTG); UTGi
(i � 1, . . ., NTG); QComi (i � 1, . . ., NCom); Tapi (i � 1, . . ., NT);
PWindx (x � 1, . . ., NW); QWindx (x � 1, . . ., NW); and LWindx
(x � 1, . . ., NW).
The decision variables are initialized within their lower
bound Zlow and upper bound Zup as shown in Section 2.
4.2. The Calculation of Dependent Variables. Before running
Mathpower program, control variables of wind turbines
including active power, reactive power, and location are
collected to calculate the new values of loads at the place-
ment of the wind turbines. Then, the data of the load must be
changed and then added in the input data of Mathpower
program. Finally, other remaining decision variables are
added to Mathpower program and running power flow for
obtaining dependent variables including PTG1; QTGi (i � 1,
. . ., NTG); ULNt (t � 1, . . ., NLN); and SBrq (q � 1, . . ., NBr).
4.3. Solution Evaluation. The quality of solution Zd is eval-
uated by calculating the fitness function. Total cost and total
active power loss are two single objectives, while the violations
of dependent variables are converted into penalty values [66].
4.4. Implementation of MEA for the Problem. In order to
reach the optimal solution for the OPF problem with the
presence of wind turbines, the implementation of MEA is
shown in the following steps and is summarized in Figure 2.

Step 1: select population N1 and maximum iteration N2.
Step 2: select and initialize decision variables for the
population as shown in Section 4.1.
Step 3: collect variables of wind turbines and tune loads.
Step 4: running Mathpower for obtaining dependent
variables shown in Section 4.2.
Step 5: evaluate quality of obtained solutions as shown
in Section 4.3.
Step 6: select the best solution Zb1 and set Iter � 1.
Step 7: select four random solutions Zr1, Zr2, Zr3, and
Zr4.
Step 8: calculate the mean fitness of the whole
population.
Step 9: produce new solutions. If Fitd < Fitmean, apply
equation (26) to produce new solution. Otherwise,
apply equation (27) to produce new solutions.
Step 10: correct new solutions using equation (23).
Step 11: collect variables of wind turbines and tune
loads.
Step 12: running Mathpower for obtaining dependent
variables shown in Section 4.2.
Step 13: evaluate quality of obtained solutions as shown
in Section 4.3.
Step 14: select good solutions among old population
and new population using equations (24) and (25).
Step 15: select the best solution Zb1.
Step 16: if Iter � N3, stop the search process and print
the optimal solution. Otherwise, set Iter to Iter + 1 and
back to Step 7.
5. Numerical Results
In this section, MEA together with four other methods
including FBI, HBO, MSGO, and CEA are applied for
placing WPPs on the IEEE 30-node system with 6 thermal
generators, 24 loads, 41 transmission lines, 4 transformers,
and 9 shunt capacitors. The ssingle line diagram of the
system is shown in Figure 3 [67]. Different study cases are
carried out as follows:
Case 1: minimization of generation cost
Case 1.1: place one wind power plant at nodes 3 and
30, respectively
Case 1.2: place one WPP at one unknown node
Case 1.3: place two wind power plants at two nodes
Case 2: minimization of power loss
Case 2.1: place one wind power plant at nodes 3 and
30, respectively
Case 2.2: place one WPP at one unknown node
Case 2.3: place two wind power plants at two nodes
The five methods are coded on Matlab 2016a and run on
a personal computer with a processor of 2.0 GHz and
4.0 GB of RAM. For each study case, fifty independent runs
are performed for each method, and the collected results
are minimum, mean, maximum, and standard deviation
values.
5.1. Electricity Generation Cost Reduction
5.1.1. Case 1.1: Place One Wind Power Plant at Nodes 3 and
30, Respectively. In this study case, one power plant is,
respectively, placed at node 3 and node 30 for comparison
of the effectiveness of the placement position. As shown in
[47], node 30 and node 3 are the most effective and in-
effective locations for placing renewable energies. The
found results from the five applied and JYA methods for the
placement of WPPs at node 3 and node 30 are reported in
Tables 2 and 3, respectively. Table 2 shows that the best cost
of MEA is $764.33, while that of other methods is from
$764.53 (CEA) to $769.963 (JYA). Exactly, MEA reaches
better cost than others by from $0.2 to $5.633. Table 3 has
the same features since MEA reaches the lowest cost of
$762.53, while that from others is from $762.62 (CEA) to
$768.039 (JYA). In addition, MEA can obtain less cost than
others by from $0.09 to $5.509. The best cost indicates that
MEA is the most powerful method among all the applied
and JYA methods, while the standard deviation (STD) of
MEA is the second lowest and it is only higher than that of
HBO.
For the effectiveness comparison between node 3 and
node 30, it concludes that node 30 is more suitable to place
WPPs. In fact, FBA, HBO, MSGO, CEA, MEA, and IAYA
[47] can reach less cost for node 30. The cost of the methods
is, respectively, $763.56, $763.24, $765.74, $762.62, $762.53,
and $768.039 for node 30, but $764.69, $764.99, $766.85,
$764.53, $764.33, and $769.963 for node 3.
Figures 4 and 5 show the best run of the applied methods
for placing WPPs at node 3 and node 30, respectively. The
curves see that MEA is much faster than the other methods
even its solution at the 70th
iteration is much better than that
of others at the final iteration.
5.1.2. Case 1.2: Place One WPP at One Unknown Node.
In this section, the location of one WPP together with active
and reactive power are determined instead of fixing the
location at node 3 and node 30 similar to Case 1.1. Table 4
indicates that MEA can reach better costs and STD than all
methods. Table 5 summarizes the results of one WPP for
Case 1.1 and Case 1.2. It is recalled that the power factor
selection of wind power plant is from 0.85 to 1.0, while the
active power is from 0 to 10 MW. For Case 1.2, HBO, CEA,
and MEA can find the same location (node 30), while FBI
and MSGO can find node 19 and node 26, respectively. Thus,
the cost of FBI and MSGO is the worst, while others have
better costs. The conclusion is very important to confirm
that the renewable energy source placement location in the
transmission power network has high impact on the effective
operation. Figure 6 shows the best run of applied methods,
and it also leads to the conclusion that MEA is the fastest
among these methods.

Start
Select population N1 and
Maximum iteration N3
Select and initialize decision variables for the
whole population section 4.1
Collect variables of wind
turbine and tune loads
Run Matpower to obtain dependent
variables shown in Section 4.2
Evaluate quality of solutions
shown in Section 4.3
Select best solution Zb1
and set Iter = 1
Select four random solutions
Zr1, Zr2, Zr3 and Zr4
Calculate mean fitness of the
whole population
Produce new solutions
following (Eq. 26)
Fitd < Fitmean
Produce new solutions
following (Eq. 27)
Correct new solutions
using (Eq. 23)
Run Mathpower to obtain dependent
variables as shown in Section 4.2
Evaluate quality of obtained solutions
as shown in Section 4.3
Select good solutions
using Eq. (24) and (Eq. 25)
Select best solution Zb1
Iter < N3
Report optimal solution
End
Iter = iter+1
Yes
No
Yes No
Figure 2: The ﬂowchart of using MEA for solving the modiﬁed OPF problem.

Figure 7 shows the fifty runs obtained by MEA. The black
curve shows fifty sorted loss values, while blue bars show the
location of the added WPP. In the figure, fifty runs are
rearranged by sorting the cost from the smallest to the
highest. Accordingly, the location of the runs is also re-
ported. The locations indicate that node 30 is found many
times, while other nodes such as 19 and 5 are also found, but
the cost of nodes 19 and 5 is much higher than that of node
30.
5.1.3. Case 1.3: Place Two Wind Power Plants at Two Nodes.
In this case, five methods are applied to minimize the cost for
two subcases, Subcase 1.3.1 with two WPPs at node 3 and
node 30 and Subcase 1.3.2 with unknown locations of two
WPPs. The results for the two cases are reported in Tables 6
and 7. MEA can reach the lowest cost for the two subcases,
which is $728.15 for Subcase 1.3.1 and $726.77 for Subcase
1.3.2. It can be seen that the locations at nodes 30 and 3 are
not as effective as locations at nodes 30 and 5. In addition to
MEA, CEA also finds the same locations at nodes 30 and 5
for Subcase 1.3.2, and CEA reaches the second-best cost
behind MEA. FBI, HBO, and MSGO cannot find the same
nodes 30 and 5, and they suffer higher cost than CEA and
MEA.
Figure 8 presents the cost and the locations of the two
WPPs obtained by 50 runs. The black curve shows fifty
values of loss sorted in the ascending order, while the blue
and orange bars show the location of the first WPP and
the second WPP. All the costs are rearranged from the
lowest to the highest values. The figure indicates that the
best cost and second-best cost are obtained by placing
WPPs at nodes 30 and 5,while the next six best costs are
obtained by placing WPPs at nodes 30 and 19. Other
worse costs are found by placing the same nodes 30 and 5
G
G
G G
G
G
1
2
3 4
5
6
7
8
9
11
10
12
13
14
15
16
18
17
20
19
21
22
24
23
25
26
27
28
29
30
Figure 3: The IEEE 30-node transmission network.
Table 2: The results obtained by methods for placing one WPP at
node 3.
Method FBI HBO MSGO CEA MEA
JYA
[47]
Minimum
cost ($/h)
764.69 764.99 766.85 764.53 764.33 769.963
Mean cost
($/h)
767.23 765.76 782.38 765.95 765.94 —
Maximum
cost ($/h)
772.62 766.71 838.51 768.83 767.91 —
STD 1.87 0.87 15.78 1.05 1.01 —
N1 10 30 15 30 30 40
N3 100 100 100 100 100 100

or nodes 30 and 19. For a few cases, two WPPs are placed
at nodes 30 and 24, but their cost is much higher. Clearly,
node 30 is the most important, and node 5 is the next
important location for supplying additional active and
reactive power.
Figures 9 and 10 show the best run of applied methods for
Subcases 1.3.1 and 1.3.2, respectively. Figure 9 shows a clear
outstanding performance of MEA over other methods since the
sixtieth iteration to the last iteration. The cost of MEA at the
sixtieth iteration is smaller than that of other methods at the
final iteration. Figure 10 also shows that MEA is much faster
than FBI, HBO, and MSGO from the 30th
iteration to the last
iteration. The cost of MEA is always smaller than these
methods from the 30th
iteration to the last iteration. CEA shows
a faster search than MEA from the first to the 80th
iteration, but
it is still worse than MEA from the 81st
iteration to the last
iteration. Obviously, MEA has a faster search than others.
5.2. Active Power Loss Reduction
5.2.1. Case 2.1: Place One Wind Power Plant at Nodes 3 and
30, Respectively. In this section, one WPP is, respectively,
located at nodes 30 and 3 for reducing power loss. Tables 8 and
9 show the obtained results from 50 trial runs. The loss of MEA
is the best for the cases of placing one WPP at node 3 and node
30. The best loss of MEA is 2.79 MW for the placement at node
3 and 2.35 MW for the placement at node 30, while those of
others are from 2.8MW to 3.339 MW for the placement at
node 3 and from 2.37MW to 2.67504MW for the placement at
node 30. On the other hand, all methods can have better loss
when placing one WPP at node 30. Clearly, node 30 needs
more supplied power than node 3. About the speed of search,
Figures 11 and 12 indicate that MEA is much more effective
than the other ones since its loss found at the 70th
iteration is
smaller than that of others at the final iteration.
5.2.2. Case 2.2. Place One WPP at One Unknown Node.
In this section, five applied methods are implemented to find
the location and power generation of one WPP. Table 10
indicates that all methods have found the same location at
node 30, but MEA is still the most effective method with the
lowest loss even it is not much smaller than others. The loss
of MEA is 2.39 MW, while that of others is from 2.45 MW to
2.7 MW. HBO is still the most stable method with the
smallest STD. Figure 13 shows the best run of five methods.
In the figure, MSGO has a premature convergence to a local
optimum with very low quality, while other methods are
searching optimal solutions. CEA seems to have a better
search process than MEA from the 1st
iteration to the 90th
iteration, but then it must adopt a higher loss from the 91st
iteration to the last iteration. Figure 14 presents the location
and the loss of the proposed MEA for 50 runs. The
Table 3: The results obtained by methods for placing one WPP at
node 30.
JYA
[47]
Minimum
cost ($/h)
763.56 763.24 765.74 762.62 762.53 768.039
Mean cost
($/h)
765.91 764.15 780.02 764.61 764.36 —
Maximum
cost ($/h)
769.50 765.82 880.75 766.62 767.88 —
STD 1.61 0.68 22.19 1.28 1.23 —
N1 10 30 15 30 30 30
N3 100 100 100 100 100 100
Fitness
value
FBI
HBO
MSGO
CEA
MEA
760
765
770
775
780
785
790
795
800
10 20 30 40 50 60 70 80 90 100
0
Iteration
Figure 4: The best runs of methods for the wind power plant
placement at node 3.
Fitness
value
FBI
HBO
MSGO
CEA
MEA
760
765
770
775
780
785
790
795
800
10 20 30 40 50 60 70 80 90 100
0
Iteration
Figure 5: The best runs of methods for the wind power plant
placement at node 30.
Table 4: The results obtained by five implemented methods for
Case 1.2.
Minimum cost ($/h) 763.96 762.72 765.22 762.89 762.52
Mean cost ($/h) 767.96 764.32 782.08 764.59 763.54
Maximum cost ($/h) 786.95 766.45 869.63 767.68 766.17
STD 4.45 0.96 18.27 1.29 0.94
N1 10 30 15 30 30
N3 100 100 100 100 100

rearranged losses from the lowest to the highest indicate that
node 30 can reduce the loss at most, while other nodes such
as 5, 7, 19, 24, and 26 are not suitable for reducing loss.
5.2.3. Case 2.2. Place Two WPPs at Two Nodes. In this
section, Subcase 2.2.1 is to place two WPPs at two pre-
determined nodes 3 and 30 and Subcase 2.2.2 is to place two
WPPs at two random nodes. Tables 11 and 12 show the
results for the two studied subcases.
The two tables reveal that MEA can reach the lowest
loss for both cases, 2.26 MW for Subcase 2.2.1 and
2.03 MW for Subcase 2.2.2. Clearly, placing WPP at the
most effective node (node 30) and the least effective node
(node 3) cannot lead to a very good solution of reducing
total loss. While, the WPP placement at node 30 and node
24 can reduce the loss from 2.26 to 2.03 MW, which is
about 0.23 MW and equivalent to 10.2%. When com-
paring to CEA, MSGO, HBO, and FBI, the proposed MEA
can save 0.02, 0.06, 0.17, and 0.11 MW for Subcase 2.2.1
and 0.02, 0.07, 0.13, and 0.21 MW for Subcase 2.2.2. The
mean loss of MEA is also smaller than that of MSGO,
HBO, and FBI and only higher than that of CEA. The STD
comparison is the same as the mean loss comparison.
Figures 15 and 16 show the search procedure of the best
run obtained by five applied methods. Figure 15 indicates
that MEA can find better parameters for wind power
plants and other electrical components than other
iteration to the last iteration.
Therefore, its loss is less than that of four remaining
to the last iteration. Figure 16
shows a better search procedure for MEA with less loss
than other ones from the 55th
iteration to the last iteration.
The two figures have the same point that the loss of MEA
at the 86th
iteration is less than that of CEA at the final
iteration. Compared to three other remaining methods,
the loss of MEA at the 67th
iteration for Subcase 2.2.1 and
at the 56th
iteration for Subcase 2.2.2 is less than that of
these methods at the final iteration. Obviously, MEA is
very strong for placing two WPPs in the IEEE 30-bus
system.
Figure 17 shows the power loss and the location of the
two WPPs for the fifty runs obtained by MEA. The black
curve shows fifty sorted loss values, while the blue and
orange bars show the location of the first WPP and the
second WPP. The view on the bars and the curve sees that
node 30 is always chosen, while the second location can
be nodes 24, 19, 21, 5, and 4. The best loss and second-
best loss are obtained at nodes 30 and 24, while other
nodes reach much higher losses.
Table 5: The optimal solutions obtained by methods for case 1.1 and case 1.2.
Case Method FBI HBO MSGO CEA MEA JYA [47]
Case 1.1 (place WPP at node 3)
Location of WPP 3 3 3 3 3 3
Generation of WPP (MW) 9.99 10.00 9.99 10.00 10.00 9.1169
Power factor of WPP 0.97 0.89 0.88 0.85 0.89 0.85
Minimum cost ($/h) 764.69 764.99 766.85 764.53 764.33 769.963
Case 1.1 (place one WPP at node 30)
Location of WPP 30 30 30 30 30 30
Minimum cost ($/h) 763.56 763.24 765.74 762.62 762.53 768.039
Case 1.2 (find the location of WPP)
Location of WPP 19 30 26 30 30 —
Generation of WPP (MW) 10.00 10.00 10.00 10.00 10.00 —
Power factor of WPP 0.93 0.97 0.92 0.99 0.95 —
Minimum cost ($/h) 763.96 762.72 765.22 762.89 762.52 —
Fitness
value
FBI
HBO
MSGO
CEA
MEA
760
765
770
775
780
785
790
795
800
10 20 30 40 50 60 70 80 90 100
0
Iteration
Figure 6: The best run of applied methods for Case 1.2.
Arranged solutions
Position
Cost ($/h)
0
5
10
15
20
25
30
Bus
Cost
($/h)
760
761
762
763
764
765
766
767
768
769
Figure 7: WPP location and cost of 50 rearranged runs obtained by
MEA.

5.3. Discussion on the Capability of MEA. In this paper, we
considered the placement of WPPs on the IEEE 30-node
system. The dimension of the system is not high, it is just
medium. In fact, among IEEE standard transmission power
systems such as IEEE 14-bus system, IEEE 30-bus system,
IEEE 57-bus system, IEEE 118-bus system, etc. The con-
sidered system is not the largest system, and it has 6 thermal
generators, 24 loads, 41 transmission lines, 4 transformers,
and 9 shunt capacitors. With the number of power plants,
lines, loads, transformers, and capacitors, the IEEE 30-bus
system is approximately as large as an area power system in a
province. By considering the placement of WPPs, the control
variables of WPPs are location, active power, and reactive
power. Therefore, there are six control variables regarding
two placed WPPs, including two locations, two values of
rated power, and two values of reactive power. In addition,
other control variables regarding optimal power flow
problem are 5 values of active power output for 6 THPs, 6
voltage values for THPs, 4 tap values for transformers, and 9
reactive power output values for shunt capacitors. On the
other hand, the dependent variables are 1 value of the active
power output for generator at slack node, 6 values of reactive
power for THPs, 41 current values of lines, and 24 voltage
values of loads. As a result, the total number of control
variables for placing two WPPs in the IEEE 30-bus system is
30, and the total number of dependent variables is 72. In the
conventional OPF problem, control variables have a high
impact on the change of dependent variables, and updating
the control variables causes the change of dependent vari-
ables. Furthermore, in the modified OPF problem, updating
the location and size of WPPs also cause the change of
Table 6: The results obtained by five implemented methods for Subcase 1.3.1.
Minimum cost ($/h) 728.61 728.40 728.80 728.39 728.15
Mean cost ($/h) 731.27 729.73 736.72 731.28 728.74
Maximum cost ($/h) 738.69 731.50 762.72 765.83 730.55
STD 1.89 0.71 7.52 5.13 0.57
N1 20 60 30 60 60
N3 100 100 100 100 100
Generation of WPP at node 30 (MW) 9.9731 10 9.9589 10 10
Generation of WPP at node 3 (MW) 9.9693 10 9.9999 9.9858 10
Power factor of WPP at node 30 0.92 0.9004 0.9803 0.9048 0.8857
Power factor of WPP at node 3 0.9782 0.9366 0.9992 0.9838 0.947
Table 7: The results obtained by five implemented methods for Subcase 1.3.2.
Minimum cost ($/h) 728.20 727.41 728.81 726.84 726.77
Mean cost ($/h) 730.67 728.69 731.75 728.17 728.04
Maximum cost ($/h) 733.77 729.89 739.08 730.59 730.35
STD 1.27 0.60 2.20 0.98 0.95
N1 20 60 30 60 60
N3 100 100 100 100 100
Locations of WPPs 30, 24 30, 19 24, 19 30, 5 30, 5
Generation of WPPs (MW) 9.95, 9.9623 10, 9.9995 10, 9.9982 10, 10 10, 10
Power factor of WPPs 0.9604, 0.9349 0.9605, 0.9394 0.9418, 0.9135 0.9342, 0.9458 0.9433, 0.8718
724
725
726
727
728
729
730
731
0
5
10
15
20
25
30
Cost
($/h)
Bus
Rearranged solutions
Position 1
Position 2
Cost ($/h)
Figure 8: WPP location and cost of 50 rearranged runs obtained by
MEA for Subcase 1.3.2.
Fitness
value
FBI
HBO
MSGO
CEA
MEA
725
730
735
740
745
20 30 40 50 60 70 80 90 100
10
Iteration
Figure 9: The best run obtained by five applied methods for
Subcase 1.3.1.

control variables such as voltage and active power of THPs.
Therefore, reaching optimal control parameters in the
modified OPF problem becomes more difficult for
metaheuristic algorithm. By experiment, MEA could solve
the conventional OPF problem successfully for the IEEE 30-
node system by setting 10 to population and 50 to iteration
Fitness
value
FBI
HBO
MSGO
CEA
MEA
720
725
730
735
740
745
750
10 20 30 40 50 60 70 80 90 100
0
Iteration
Figure 10: The best run obtained by five applied methods for Subcase 1.3.2.
Table 8: The results obtained by methods as placing one WPP at node 3 for loss reduction.
Method FBI HBO MSGO CEA MEA JYA [47]
Minimum power loss (MW) 2.91 2.87 2.91 2.80 2.79 3.3390
Mean power loss (MW) 3.35 3.08 4.01 3.11 3.10 —
Maximum power loss (MW) 4.68 3.30 6.87 3.40 3.36 —
STD 0.39 0.10 1.09 0.16 0.15 —
Power factor of WPP 0.9601 0.9989 0.974 0.9168 1 0.85
Table 9: The results obtained by methods as placing one WPP at node 30 for loss reduction.
Minimum power loss (MW) 2.51 2.47 2.43 2.37 2.35 2.67504
Mean power loss (MW) 2.96 2.62 3.23 2.73 2.65 —
Maximum power loss (MW) 4.50 2.85 5.69 3.18 2.97 —
STD 0.43 0.09 0.89 0.17 0.14 —
N1 10 30 15 30 30 30
N3 100 100 100 100 100 100
Power
loss
(MW)
FBI
HBO
MSGO
CEA
MEA
3
5
10 20 30 40 50 60 70 80 90 100
0
Iteration
3.5
4
4.5
Figure 11: The best runs of methods for minimizing loss as placing
one WPP at node 3.
Fitness
value
FBI
HBO
MSGO
CEA
MEA
2
4
10 20 30 40 50 60 70 80 90 100
0
Iteration
2.5
3
3.5
Figure 12: The best runs of methods for minimizing loss as placing
one WPP at node 30.

number and MEA could reach the most optimal solutions by
setting 15 to population and 75 to iteration number.
However, for the modified problem with the placement of
two WPPs, the settings to reach the best performance for
MEA were 60 for population and 100 for the iteration
number. Clearly, the setting values were higher for the
modified OPF problem. About the average simulation time
for the study cases, Table 13 summarizes the time from all
methods for all study cases. Comparisons of the computa-
tion time indicate that MEA has the same computation time
as FBI, HBO, MSGO, and CEA, but it has shorter time than
JYA [47]. The average time for applied methods is about 30
seconds for the cases of placing one WPP and about 53
seconds for other cases of placing two WPPs, while the time
is about 72 seconds for JYA for the cases of placing one WPP.
The five algorithms approximately have the same average
time because the setting of population and iteration number
is the same. The reported time of the proposed method is not
too long for a system with 30 nodes, and it seems that MEA
can be capable for handling a real power system or a larger-
scale power system. Therefore, we have tried to apply MEA
for other larger scale systems with 57 or 118 nodes. For
conventional OPF problem without the optimal placement
of WPPs, MEA could solve the conventional OPF problem
successfully. However, for the placement of WPPs in
modified OPF problem for the IEEE 57-node system and the
IEEE 118-node system, MEA could not succeed to reach
valid solutions. Therefore, the highest shortcoming of the
Fitness
value
FBI
HBO
MSGO
CEA
MEA
2
4
10 20 30 40 50 60 70 80 90 100
0
Iteration
2.5
3
3.5
Figure 13: The best runs of five applied methods for Subcase 2.2.
0.5
1
1.5
2
2.5
3
3.5
Position
Power loss (MW)
0
5
10
15
20
25
30
Bus
Power
loss
(MW)
0
4
Figure 14: WPP location and loss of 50 rearranged runs obtained by MEA for Subcase 2.2.
Table 10: The results obtained by methods for placing one WPP at one unknown node.
Minimum power loss (MW) 2.70 2.49 2.46 2.45 2.39
Mean power loss (MW) 3.17 2.68 3.54 2.82 2.74
Maximum power loss (MW) 4.72 3.18 6.34 3.29 3.43
STD 0.40 0.13 0.97 0.21 0.21
N1 10 30 15 30 30
N3 100 100 100 100 100
Found position 30 30 30 30 30
Generation of WPP (MW) 8.8911 9.9998 9.9065 9.9855 9.9956
Power factor of WPP 0.9899 0.9503 0.9913 0.895 0.9122

Fitness
value
FBI
HBO
MSGO
CEA
MEA
2
4
10 20 30 40 50 60 70 80 90 100
0
Iteration
2.2
2.4
2.6
2.8
3
3.2
3.4
3.6
3.8
Figure 16: The best runs obtained by methods for Subcase 2.2.2.
Table 11: The results obtained by ﬁve methods for Subcase 2.2.1.
STD 0.14 0.09 0.25 0.13 0.13
N1 20 60 30 60 60
N3 100 100 100 100 100
Location of WPPs 3, 30 3, 30 3, 30 3, 30 3, 30
Generation of WPPs (MW) 9.6108, 9.8128 7.4719, 9.9936 6.8908, 9.9674 9.8404, 9.9999 9.9922, 9.9299
Table 12: The results obtained by ﬁve methods for Subcase 2.2.2.
STD 0.28 0.13 0.37 0.15 0.14
N1 20 60 30 60 60
N3 100 100 100 100 100
Locations of WPPs 30, 19 19, 30 30, 19 30, 19 24, 30
Generation of WPPs (MW) 9.83, 9.28 9.98, 9.97 9.70, 8.15 9.99, 9.98 9.95, 9.99
Fitness
value
FBI
HBO
MSGO
CEA
MEA
2
4
10 20 30 40 50 60 70 80 90 100
0
Iteration
2.2
2.4
2.6
2.8
3
3.2
3.4
3.6
3.8
Figure 15: The best runs obtained by methods for Subcase 2.2.1.

study is not to reach the successful application of MEA for
placing WPPs on large-scale systems with 57 and 118 nodes.
It can be stated that CEA and MEA are powerful op-
timization tools for the IEEE 30-node system, but their
capability on other large-scale systems or real systems is
limited. The methods may need more effective improvement
to overcome the mentioned limitation.
6. Conclusions
In this paper, a modified OPF (MOPF) problem with the
placement of wind power plants in an IEEE 30-bus trans-
mission power network was solved by implementing four
conventional metaheuristic algorithms and the proposed
MEA. Two single objectives taken into account were min-
imization of total generation cost and minimization of
power loss. About the number of WPPs located in the
system, two cases are, respectively, one WPP and two WPPs.
About the locations of the WPPs, simple cases were to accept
the result from the previous study [47]. Buses 30 and 3 were
the most effective and ineffective locations. The results in-
dicated that the placement of one WPP at bus 30 can reach
smaller power loss and smaller fuel cost than at bus 3. For
other complicated cases, the paper also investigated the
effectiveness of locations by applying MEA and four other
metaheuristic algorithms to determine the locations. As a
result, placing one WPP at bus 30 has reached the smallest
power loss and the smallest total fuel cost. For placing two
WPPs, buses 30 and 3 could not result in the smallest fuel
cost and the smallest power loss. Buses 30 and 5 were the best
locations for the minimization of fuel cost, while buses 30
and 24 were the best locations for the minimization of power
loss. Therefore, the main contribution of the study regarding
the electrical field is to determine the best locations for the
best power loss and the best total cost.
For placing one WPP, fuel costs of MEA were the smallest
and equal to $764.33 and $762.53 for locations at node 3 and
node 30, whilst those of others were much higher and equal to
$769.963 and $768.039, respectively. For placing two WPPs at
two found locations, MEA has reached the cost of $726.77, but
the worst cost of others was $728.81. The power losses of MEA
were also reduced significantly as compared to others. For
placing one WPP at node 3 and node 30, MEA has reached 2.79
and 2.35MW, but those of others have been larger and equal to
3.339 and 2.67504MW, respectively. For placing two WPPs at
two found locations, the best loss of 2.03MW was found by
MEA and the worst loss of 2.24MW was found by others. In
summary, the proposed MEA could attain lesser cost than
others from 0.28% to 0.73% and lesser power loss than others
from 9.38% to 16.44%. Clearly, the improvement levels are
significant. However, for other systems with larger scale, MEA
could not succeed in determining the best location and size for
WPPs. Thus, in the future work, we will find solutions to
improve MEA for larger systems and real systems. In addition,
we will also consider more renewable energy power plants,
such as photovoltaic power plants and uncertainty charac-
teristics of solar and wind speed. All considered complexities
will form a real problem as a real power system, and contri-
butions of optimization algorithms and renewable energies will
be shown clearly.
Table 13: Average computation time of each run obtained by methods for study cases.
Case 1.1 (place one WPP at node 3) 35.21 30.65 32.82 28.18 28.45 ∼72.4
Case 1.2 (place one WPP at one unknown node) 27.93 31.92 27.32 27.64 27.76 —
Subcase 1.3.1 (place two WPPs at nodes 3 and 30) 51.04 53.36 53.56 53.38 52.57 —
Subcase 1.3.2 (place two WPPs at two unknown nodes) 55.18 54.05 56.27 54.18 55.14 —
Case 2.2 (place one WPP at one unknown node) 28.20 26.75 28.27 29.62 27.40 —
Subcase 2.2.1 (place two WPPs at nodes 3 and 30) 55.41 56.43 56.43 56.81 56.09 —
Subcase 2.2.2 (place one WPP at one unknown node) 54.04 54.14 59.2 55.19 56.12 —
1
4
0
5
10
15
20
25
30
Power
loss
(MW)
Bus
Position 1
Position 2
Power loss (MW)
1.5
2
2.5
3
3.5
Figure 17: Power losses and locations of two WPPs for 50 runs obtained by MEA.

Nomenclature
Fitmean: Mean fitness of the available population
Iq, Rq: Current and resistance of the qth
transmission line
NTG: Quantity of thermal units
Nn: Quantity of nodes in networks
NLN, NBr: Quantity of load nodes and
transmission lines
NT: Quantity of transformers
N1: Population
N2: Number of control decision variables
N3: The maximum iteration
NW: Number of nodes with the presence of
wind turbines
PTGi, QTGi: Active and reactive power generation of
the ith generator
Prqi: Active power required by load at the ith
node
Pmin
TGi, Pmax
TGi : Minimum and maximum output of
active power generated by the ith
generator
NCom: Number of nodes with shunt capacitors
Qmin
TGi, Qmax
TGi : Minimum and maximum output of
reactive power generated by the ith
generator
QTGx: Reactive power generated by the
thermal generator located at node x
Qrqx: Reactive power required by load at node
x
QComx: Reactive power supplied by the
compensator that actually is capacitor
bank placed at node x
Qmin
Comx, Qmax
Comx: Minimum and maximum output of
reactive power generated by shunt
capacitors at node x
Pmin
Wind, Pmax
Wind,
Qmin
Wind, Qmax
Wind:
Minimum and maximum output of
active and reactive power generated by
wind turbines
r1, r2, r3, r4, r5, r6,
r7, r8:
Random number within 0 and 1
SBrq: Operating apparent power of the qth
line
Smax
Br : Maximum limit of apparent power of
the qth line
Tapi: Selected tap of the transformer i
Tapmin
, Tapmax
: The lowest and highest tap settings of
transformers
Ux, Uy: The voltage magnitudes at node x and
node y
UTGi: Voltage of the ith generator
Umin
TGi, Umax
TGi : Minimum and maximum voltage of the
ith generator
ULNt: Operating voltage of the tth load
Umin
LN , Umax
LN : Minimum and maximum operating
voltage of loads
Yxy: The admittance value between node x
and node y
μ1, μ2, μ3: Electricity generation cost coefficient
φx, φx: The voltage phasors at node x and node
y
Zr1, Zr2, Zr3, Zr4: Randomly chosen solutions from
population
AFAPA: Adaptive fuzzy artificial physics
algorithm
AA: Antlion algorithm
APDE: Adaptive parameters based differential
evolution
ACA: Ant colony algorithm
ABA: Artificial bee algorithm
APO: Artificial physics algorithm
ALPOS: Aging leader-based particle swarm
optimization
BFA: Bacteria foraging algorithm
BSA: Bird swarm algorithm
BWOA: Black widow optimization algorithm
BMA: Barnacles mating algorithm
BA: Bat algorithm
BSA: Backtracking search algorithm
CSA1: Crow search algorithm
CSA: Cuckoo search algorithm
DE: Differential evolution
EGC: Electricity generation cost
FPA: Flower pollination algorithm
GSO: Glowworm swarm algorithm
GWA: Grey wolf algorithm
GSA: Gravitational search algorithm
GA: Genetic algorithm
HABC: Hybrid artificial bee colony algorithm
HA: Hybrid algorithm
HGTPEA: Hybrid genetic and two-point
estimation algorithm
HGNIPA: Hybrid genetic and nonlinear interior
point algorithm
HSQTIICA: Hybrid sequential quadratic technique
and improved imperialist competitive
algorithm
IMVA: Improved multiverse algorithm
IMA: Ion motion algorithm
IICA: Improved imperialist competitive
algorithm
JYA: JAYA algorithm
KHA: Krill herd algorithm
MBFA: Modified bacteria foraging algorithm
MCS: Modified cuckoo search
MFO: Moth flame optimization
MHGSPSO: Modified hybrid gravitational search
algorithm and particle swarm
optimization
MVA: Multiverse algorithm
MFA: Moth flame algorithm
MJYA: Modified JAYA
MPSO: Modified particle swarm optimization
MDE: Modified differential evolution
MSA: Moth swarm algorithm
NSGA-II: Improved nondominated sorting
genetic algorithm

NDSGWA: Nondominated sort grey wolf algorithm
PSO: Particle swarm optimization
RCGA: Real coded genetic algorithm
TUs: Thermal units
WTs: Wind turbines
WA: Whale algorithm
SSA: Salp swarm algorithm.
Data Availability
The input data for the IEEE 30-bus system in this study are
available from the literature.
Conflicts of Interest
The authors declare that they have no conflicts of interest.
Acknowledgments
This research is funded by Vietnam National Foundation for
Science and Technology Development (NAFOSTED) under
grant number. 102.02-2020.07.
References
[1] L. Le Dinh, D. Vo Ngoc, and P. Vasant, “Artificial bee colony
algorithm for solving optimal power flow problem,” The
Science World Journal, vol. 2013, Article ID 159040, 9 pages,
2013.
[2] L. Sun, J. Hu, and H. Chen, “Artificial bee colony algorithm
based on K-means clustering for multiobjective optimal
power flow problem,” Mathematical Problems in Engineering,
vol. 2015, Article ID 762853, 18 pages, 2015.
[3] R. Roy and H. T. Jadhav, “Optimal power flow solution of
power system incorporating stochastic wind power using
Gbest guided artificial bee colony algorithm,” International
Journal of Electrical Power & Energy Systems, vol. 64,
pp. 562–578, 2015.
[4] G. Chen, S. Qiu, Z. Zhang, Z. Sun, and H. Liao, “Optimal
power flow using gbest-guided cuckoo search algorithm with
feedback control strategy and constraint domination rule,”
Mathematical Problems in Engineering, vol. 201714 pages,
Article ID 9067520, 2017.
[5] T. L. Duong, N. A. Nguyen, and T. T. Nguyen, “A newly
hybrid method based on cuckoo search and sunflower op-
timization for optimal power flow problem,” Sustainability,
vol. 12, no. 13, p. 5238, 2020.
[6] P. Umapathy, C. Venkataseshaiah, and M. S. Arumugam,
“Particle swarm optimization with various inertia weight
variants for optimal power flow solution,” Discrete Dynamics
in Nature and Society, vol. 2010, Article ID 462145, 15 pages,
2010.
[7] P. K. Roy, S. P. Ghoshal, and S. S. Thakur, “Multi-objective
optimal power flow using biogeography-based optimization,”
Electric Power Components and Systems, vol. 38, no. 12,
pp. 1406–1426, 2010.
[8] A.-F. Attia, Y. A. Al-Turki, and A. M. Abusorrah, “Optimal
power flow using adapted genetic algorithm with adjusting
population size,” Electric Power Components and Systems,
vol. 40, no. 11, pp. 1285–1299, 2012.
[9] R. Arul, G. Ravi, and S. Velusami, “Solving optimal power
flow problems using chaotic self-adaptive differential
harmony search algorithm,” Electric Power Components and
Systems, vol. 41, no. 8, pp. 782–805, 2013.
[10] H. R. E.-H. Bouchekara and M. A. Abido, “Optimal power
flow using differential search algorithm,” Electric Power
Components and Systems, vol. 42, no. 15, pp. 1683–1699, 2014.
[11] H. R. El-Hana Bouchekara, M. A. Abido, and A. E. Chaib,
“Optimal power flow using an improved electromagnetism-
like mechanism method,” Electric Power Components and
Systems, vol. 44, no. 4, pp. 434–449, 2016.
[12] M. A. Taher, S. Kamel, F. Jurado, and M. Ebeed, “An improved
moth-flame optimization algorithm for solving optimal
power flow problem,” International Transactions on Electrical
Energy Systems, vol. 29, no. 3, Article ID e2743, 2019.
[13] M. A. Taher, S. Kamel, F. Jurado, and M. Ebeed, “Modified
grasshopper optimization framework for optimal power flow
solution,” Electrical Engineering, vol. 101, no. 1, pp. 121–148,
Article ID 0123456789, 2019.
[14] M. Pourakbari-Kasmaei, M. Lehtonen, M. Fotuhi-Firuzabad,
M. Marzband, and J. R. S. Mantovani, “Optimal power flow
problem considering multiple-fuel options and disjoint op-
erating zones: a solver-friendly MINLP model,” International
Journal of Electrical Power & Energy Systems, vol. 113,
pp. 45–55, 2019.
[15] K. Srilakshmi, P. Ravi Babu, and P. Aravindhababu, “An
enhanced most valuable player algorithm based optimal
power flow using Broyden’s method,” Sustainable Energy
Technologies and Assessments, vol. 42, Article ID 100801, 2020.
[16] K. Nusair and F. Alasali, “Optimal power flow management
system for a power network with stochastic renewable energy
resources using golden ratio optimization method,” Energies,
vol. 13, no. 14, p. 3671, 2020.
[17] M. Z. Islam, N. I. A. Wahab, V. Veerasamy et al., “A harris
hawks optimization based single- and multi-objective optimal
power flow considering environmental emission,” Sustain-
ability, vol. 12, no. 13, p. 5248, 2020.
[18] S. Li, W. Gong, L. Wang, X. Yan, and C. Hu, “Optimal power
flow by means of improved adaptive differential evolution,”
Energy, vol. 198, Article ID 117314, 2020.
[19] E. Naderi, M. Pourakbari-Kasmaei, F. V. Cerna, and
M. Lehtonen, “A novel hybrid self-adaptive heuristic algo-
rithm to handle single- and multi-objective optimal power
flow problems,” International Journal of Electrical Power &
Energy Systems, vol. 125, Article ID 106492, 2021.
[20] M. Z. Islam, N. I. Abdul Wahab, V. Veerasamy et al.,
“Generation fuel cost and loss minimization using salp swarm
algorithm based optimal power flow,” in Proceedings of the
2020 International Conference on Computer Communication
and Informatics, pp. 1–6, Coimbatore, India, January 2020.
[21] P. P. Biswas, P. N. Suganthan, R. Mallipeddi, and
G. A. J. Amaratunga, “Optimal power flow solutions using
differential evolution algorithm integrated with effective
constraint handling techniques,” Engineering Applications of
Artificial Intelligence, vol. 68, pp. 81–100, 2018.
[22] G. Chen, Z. Lu, and Z. Zhang, “Improved krill herd algorithm
with novel constraint handling method for solving optimal
power flow problems,” Energies, vol. 11, no. 1, p. 76, 2018.
[23] A. Khelifi, B. Bentouati, and S. Chettih, “Optimal power flow
problem solution based on hybrid firefly krill herd method,”
International Journal of Engineering Research in Africa,
vol. 44, pp. 213–228, 2019.
[24] A. Alhejji, M. Ebeed Hussein, S. Kamel, and S. Alyami,
“Optimal power flow solution with an embedded center-node
unified power flow controller using an adaptive grasshopper

optimization algorithm,” IEEE Access, vol. 8, pp. 119020–
119037, 2020.
[25] A. Panda and M. Tripathy, “Optimal power flow solution of
wind integrated power system using modified bacteria for-
aging algorithm,” International Journal of Electrical Power &
Energy Systems, vol. 54, pp. 306–314, 2014.
[26] A. Panda and M. Tripathy, “Security constrained optimal
power flow solution of wind-thermal generation system using
modified bacteria foraging algorithm,” Energy, vol. 93,
pp. 816–827, 2015.
[27] X. Hao, C. Jiang, L. Wu, and L. Zhang, “Based on the power
factors of dfig wind farm for power flow optimization,”
Proceedings of the Second International Conference on
Mechatronics and Automatic Control, vol. 334, pp. 157–165,
2015.
[28] C. Mishra, S. P. Singh, and J. Rokadia, “Optimal power flow in
the presence of wind power using modified cuckoo search,”
IET Generation, Transmission & Distribution, vol. 9, no. 7,
pp. 615–626, 2015.
[29] A. Panda and M. Tripathy, “Solution of wind integrated
thermal generation system for environmental optimal power
flow using hybrid algorithm,” Journal of Electrical Systems and
Information Technology, vol. 3, no. 2, pp. 151–160, 2016.
[30] S. B. Pandya and H. R. Jariwala, “Stochastic wind-thermal
power plants integrated multi-objective optimal power flow,”
Majlesi Journal of Electrical Engineering, vol. 14, no. 2,
pp. 93–110, 2020.
[31] K. Teeparthi and D. M. Vinod Kumar, “Security-constrained
optimal power flow with wind and thermal power generators
using fuzzy adaptive artificial physics optimization algo-
rithm,” Neural Computing and Applications, vol. 29, no. 3,
pp. 855–871, 2018.
[32] A. Abdollahi, A. Ghadimi, M. Miveh, F. Mohammadi, and
F. Jurado, “Optimal power flow incorporating facts devices
and stochastic wind power generation using krill herd al-
gorithm,” Electronics, vol. 9, no. 6, p. 1043, 2020.
[33] S. R. Salkuti, “Optimal power flow using multi-objective
glowworm swarm optimization algorithm in a wind energy
integrated power system,” International Journal of Green
Energy, vol. 16, no. 15, pp. 1547–1561, 2019.
[34] S. Duman, J. Li, L. Wu, and U. Guvenc, “Optimal power flow
with stochastic wind power and FACTS devices: a modified
hybrid PSOGSA with chaotic maps approach,” Neural
Computing and Applications, vol. 32, no. 12, pp. 1–30, 2020.
[35] M. Ahmad, N. Javaid, I. A. Niaz, S. Shafiq, O. Ur Rehman, and
H. Majid Hussain, Application of Bird Swarm Algorithm for
Solution of Optimal Power Flow Problems, Springer Inter-
national Publishing, Berlin, Germany, 2019.
[36] M. Abdullah, N. Javaid, A. Chand, and Z. A. Khan, Multi-
objective Optimal Power Flow Using Improved Multi-Objective
Multi-Verse Algorithm, Springer International Publishing,
Berlin, Germany, 2019.
[37] T. Samakpong, W. Ongsakul, and M. Nimal Madhu, “Optimal
power flow considering cost of wind and solar power un-
certainty using particle swarm optimization,” Advances in
Intelligent Systems and Computing, vol. 1072, pp. 190–203,
2020.
[38] E. X. S. Araujo, M. C. Cerbantes, and J. R. S. Mantovani,
“Optimal power flow with renewable generation: a modified
NSGA-II-based probabilistic solution approach,” Journal of
Control, Automation and Electrical Systems, vol. 31, no. 4,
pp. 979–989, 2020.
[39] S. B. Pandya and H. R. Jariwala, “Stochastic renewable energy
resources integrated multi-objective optimal power flow,”
TELKOMNIKA (Telecommunication Computing Electronics
and Control), vol. 18, no. 3, pp. 1582–1599, 2020.
[40] I. U. Khan, N. Javaid, K. A. A. Gamage, C. J. Taylor, S. Baig,
and X. Ma, “Heuristic algorithm based optimal power flow
model incorporating stochastic renewable energy sources,”
IEEE Access, vol. 8, pp. 148622–148643, 2020.
[41] M. H. Sulaiman and Z. Mustaffa, “Optimal power flow in-
corporating stochastic wind and solar generation by meta-
heuristic optimizers,” Microsystem Technologies, vol. 27, no. 9,
pp. 1–15, 2020.
[42] F. Daqaq, M. Ouassaid, R. Ellaia, and A. T. Zeine, “Optimal
power flow solution including stochastic renewable re-
sources,” in Proceedings of the 2018 6th International Re-
newable and Sustainable Energy Conference (IRSEC), pp. 1–6,
Rabat, Morocco, December 2018.
[43] P. P. Biswas, P. N. Suganthan, and G. A. J. Amaratunga,
“Optimal power flow solutions incorporating stochastic wind
and solar power,” Energy Conversion and Management,
vol. 148, pp. 1194–1207, 2017.
[44] S. S. Reddy, “Optimal power flow with renewable energy
resources including storage,” Electrical Engineering, vol. 99,
no. 2, pp. 685–695, 2017.
[45] J. Jiang, X. Han, J. Wang, X. Zhu, D. Sun, and Y. Ma, “Optimal
power flow with transmission switching for power system
with wind/photovoltaic generation,” in Proceedings of the
2017 Chinese Automation Congress, CA, pp. 5802–5806, Jinan,
China, October 2017.
[46] S. B. Pandya and H. R. Jariwala, “Renewable energy resources
integrated multi-objective optimal power flow using non-
dominated sort grey wolf optimizer,” Journal of Green En-
gineering, vol. 10, no. 1, pp. 180–205, 2020.
[47] W. Warid, H. Hizam, N. Mariun, and N. Abdul-Wahab,
“Optimal power flow using the Jaya algorithm,” Energies,
vol. 9, no. 9, p. 678, 2016.
[48] J. Ben Hmida, T. Chambers, and J. Lee, “Solving constrained
optimal power flow with renewables using hybrid modified
imperialist competitive algorithm and sequential quadratic
programming,” Electric Power Systems Research, vol. 177,
Article ID 105989, 2019.
[49] E. E. Elattar and S. K. ElSayed, “Modified JAYA algorithm for
optimal power flow incorporating renewable energy sources
considering the cost, emission, power loss and voltage profile
improvement,” Energy, vol. 178, pp. 598–609, 2019.
[50] M. Zhang and Y. Li, “Multi-objective optimal reactive power
dispatch of power systems by combining classification-based
Multi-objective evolutionary algorithm and integrated deci-
sion making,” IEEE Access, vol. 8, pp. 38198–38209, 2020.
[51] V. Suresh and S. S. Kumar, “Optimal reactive power dispatch
for minimization of real power loss using SBDE and DE-
strategy algorithm,” Journal of Ambient Intelligence and
Humanized Computing, pp. 1–15, 2020, In press.
[52] Y. Li and Y. Li, “Security-constrained multi-objective optimal
power flow for a hybrid AC/VSC-MTDC system with lasso-
based contingency filtering,” IEEE Access, vol. 8, pp. 6801–
6811, 2018.
[53] Y. Li, Y. Li, G. Li, D. Zhao, and C. Chen, “Two-stage multi-
objective OPF for AC/DC grids with VSC-HVDC: incorpo-
rating decisions analysis into optimization process,” Energy,
vol. 147, pp. 286–296, 2018.
[54] A. Faramarzi, M. Heidarinejad, B. Stephens, and S. Mirjalili,
“Equilibrium optimizer: a novel optimization algorithm,”
Knowledge-Based Systems, vol. 191, Article ID 105190, 2020.
[55] D. T. Abdul-hamied, A. M. Shaheen, W. A. Salem, W. I. Gabr,
and R. A. El-sehiemy, “Equilibrium optimizer based multi

dimensions operation of hybrid AC/DC grids,” Alexandria
Engineering Journal, vol. 59, no. 6, pp. 4787–4803, 2020.
[56] A. M. Shaheen, A. M. Elsayed, R. A. El-Sehiemy, and
A. Y. Abdelaziz, “Equilibrium optimization algorithm for
network reconfiguration and distributed generation alloca-
tion in power systems,” Applied Soft Computing, vol. 98,
Article ID 106867, 2021.
[57] H. Ozkaya, M. Yıldız, A. R. Yıldız, S. Bureerat, B. S. Yıldız, and
S. M. Sait, “The equilibrium optimization algorithm and the
response surface-based metamodel for optimal structural de-
sign of vehicle components,” Materials Testing, vol. 62, no. 5,
pp. 492–496, 2020.
[58] A. B. Krishna, S. Saxena, and V. K. Kamboj, “A novel sta-
tistical approach to numerical and multidisciplinary design
optimization problems using pattern search inspired Harris
hawks optimizer,” Neural Computing and Applications,
vol. 33, no. 12, pp. 7031–7072, 2021.
[59] M. Abdel-Basset, R. Mohamed, S. Mirjalili,
R. K. Chakrabortty, and M. J. Ryan, “Solar photovoltaic pa-
rameter estimation using an improved equilibrium opti-
mizer,” Solar Energy, vol. 209, pp. 694–708, 2020.
[60] J. Zhao and Z. Gao, “The improved equilibrium optimization
algorithm with best candidates,” Journal of Physics: Confer-
ence Series, vol. 1575, no. 1, Article ID 012089, 2020.
[61] J. Zhao and Z. M. Gao, “The improved mayfly optimization
algorithm with Chebyshev map,” Journal of Physics: Confer-
ence Series, vol. 1684, pp. 343–346, 2020.
[62] A. Wunnava, M. K. Naik, R. Panda, B. Jena, and A. Abraham,
“A novel interdependence based multilevel thresholding
technique using adaptive equilibrium optimizer,” Engineering
Applications of Artificial Intelligence, vol. 94, Article ID
103836, 2020.
[63] Q. Askari, M. Saeed, and I. Younas, “Heap-based optimizer
inspired by corporate rank hierarchy for global optimization,”
Expert Systems with Applications, vol. 161, Article ID 113702,
2020.
[64] J.-S. Chou and N.-M. Nguyen, “FBI inspired meta-optimi-
zation,” Applied Soft Computing, vol. 93, Article ID 106339,
2020.
[65] A. Naik, S. C. Satapathy, and A. Abraham, “Modified social
group optimization-a meta-heuristic algorithm to solve short-
term hydrothermal scheduling,” Applied Soft Computing,
vol. 95, p. 106524, 2020.
[66] T. T. Nguyen, “A high performance social spider optimization
algorithm for optimal power flow solution with single ob-
jective optimization,” Energy, vol. 171, pp. 218–240, 2019.
[67] R. D. Zimmerman and C. E. Murillo-Sánchez: http://www.
pserc.cornell.edu/matpower, 2017.

1015367.pdf

Recommended

Recommended

More Related Content

Similar to 1015367.pdf

Similar to 1015367.pdf (20)

More from TesloachKongGilo

More from TesloachKongGilo (10)

Recently uploaded

Recently uploaded (20)

1015367.pdf